 In episode three, we touched on multiple independent error factors. Now, imagine you're applying for a job. You've got a job interview. Now on the particular day, there's going to be some things working for you and some things working against you. Now imagine if you had a migraine, or think about this, an hour or two before your interview, your pet dog died, right? That's not good. There are going to be some things working for you. You might have had an awesome night's sleep. The sun is shining. You just feel amazing, right? These are the things that are going to have a really large impact on how well you impress that interview panel and whether you get the job or not. So, number one, they're multiple. So all of these factors, there's lots of them and they're easy to ignore. And statistically, we call this noise. Number two, they're independent. So the relationship between your dog dying and the sun shining is probably not linked. Those things are fairly confident that they're independent. So these factors are going to have a large impact on how well you perform in that interview. That's right. And it comes down to the fact that on any sort of event, some things are going to work in your favor. It feels, and again, imagine, sometimes it feels like that, right? Things are conspiring against you in the world. It feels like the world is ganging up on you and everything is just going really badly and maybe even the source of depression. But on the other hand, some things it just feels, you're on the top of your game, you have soundtrack playing in your head, you're walking around, it's fantastic. Everything you touch turns to gold. And as we've just seen in the last episode, these random events are lumpy. Sometimes these things, sometimes there are runs of good things, sometimes there's runs of bad things, but they end. And I think that idea, the fact that chance is lumpy, the fact that these multiple independent error factors are operating, it's related to this idea of something called regression to the mean, or Sir Francis Galton called it regression to mediocrity. And that's exactly, the observable thing is exactly as we've described. In the long run, things tend to balance out. These lumpy things that happen initially tend to smooth out later on. And that's really what we're talking about here with regression to the mean. We see it all the time in sport. You have the star athlete who performs exceptionally well, as Tom Gilovich said, they end up on the cover of sports illustrated or the Madden NFL game, the front of the video game cover. And there are all sorts of superstitions, jinxes that surround these phenomena. So that star athlete doesn't perform as well after being featured on the cover. Now that has nothing to do with any sort of superstition or magical belief. It just has everything to do with this idea of regression to the mean. That lump smoothed out in the long run and their behavior kind of tapered off. But I think it's important that we make this idea of regression to the mean a bit more concrete. So let's take an example. Imagine that we have a class, a classroom. People sitting at their desks, 100 students. They're all sitting at their desks and we're going to give them an exam. The exam has 100 questions on it, all true and false. So each question on the exam is either true or it's false. Now, all 100 people who are sitting in that classroom, not a single one of them speak a word of English. They're completely, completely novices when it comes to speaking the English language. And so they don't understand a single question on that entire exam. So they're just reading this gibberish and then marking either true or false as a response. Now, as a result, so we have 100 people, 100 questions. As a result of taking that test, you're going to have some people who took that test. Most people are going to get about 50% on average, but some people are going to perform better than 50%. In fact, with a group of about 100 people, I would imagine to have two or even three people who score 60% or even higher, say 65%. So say we have this top performer, Billy, in this test, took the exam, doesn't understand a word of it, but he scored 65%. Just by chance, just completely randomly. We have another student say Jane. She scored on the other end, so 35% correct. So we put Billy over here, the top performer, Jane over here, she's the bottom performer. Now what we want people to do is imagine what happens on day two. Day two, we perform exactly the same test, a different test, different questions, though it doesn't really matter, they're guessing anyway, but we sit them down at their desk to take another exam and ask them to mark true or false. What's going to happen on day two? How would you expect Billy to perform on day two and how would you expect Jane to perform on day two? So we'll ask the students to go ahead and make their predictions before we move on.